3356
J. Am. Chem. SOC.1993,115, 3356-3357
Protonated 3-Fluoropiperidines: An Unusual Fluoro Directing Effect and a Test for Quantitative Theories of Solvation1
HJ&
H
co**
David C. Lankin,' Nizal S. Chandrakumar, Shashidhar N. Rao, Dale P. Spangler, and James P. Snyder'-+
Searle R and D, 4901 Searle Parkway Skokie, Illinois 60077 Received September 14, 1992 The ability to predict the energetics of molecules immersed in solvent is the principle that unites the theory and practice of solution chemistry. For neutral, polar, and some singly charged species, the resource-demanding free energy perturbation method with explicit treatment of solvent has yielded impressive results.2 Solvent implicit parametric methods based on generalized Born electrostatic^,^^^ reaction fields5 and molecular surfaces6 are likewise capable of quantitative or semiquantitative assessment of solvated species. Exceptionally demanding are cases of highly polar or multiply charged compounds with conformational or fluxional freedom in the condensed phase. Many drugs at physiological pH,' ionic aggregates, metal complexes, and zwitterions sample the range of possibilities. To provide a test for current and emerging solvent-based theoretical methodology, we have prepared a series of singly and doubly charged (zwitterionic) piperidines and analyzed their conformational profiles in water by NMR. An unprecedented fluoro-directed conformational effect has been observed. In addition, series 1 - 8 has been subjected to evaluation by several aqueous solvation treatments. The conformational profiles of piperidines 1 - 4 in water (D20) were determined by 1- and 2-D NMR. A summary of conformer populations as derived by NMR and modeling is given in Table I. The proton NMR spectrum of 3-carboxypiperidine (nipecotic acid), 1, in D20 (pH ca. 6, zwitterion, * = -) exhibits couplings between H-2 and H-3 (3J~.2axl~.3 = 9.4 Hz and 3 J ~ . 2 , / ~ .=3 3.9 Hz) that demonstrate a 65-75% population of the equatorial conformer consistent with earlier studies8using theJ-value method as applied t~piperidines.~ Current address: Department of Medicinal Chemistry, Istituto Ricerche Biologia Molecolari (IRBM), Via Pontina Km 30.600,0oO40Pomezia (Roma), Italy. (1) A preliminary account of this work was presented at the 33rd Experimental NMR Conference (ENC), Asilomar, CA, March 29-April 2, 1992, Poster WP-34. (2) (a) Computer Simulation of Biomolecular Systems; Van Gunsteren, W., Weiner, P. K., Eds.; ESCOM: Leiden, 1989. (b) Jorgensen, W. L. Acc. Chem. Res. 1989, 22, 184. (c) Kollman, P. A,; Merz, K. M., Jr. Acc. Chem. Res. 1990, 23, 246-252. (3) Still, C. W.; Tempczyk, A.; Hawley, R. C.; Hendrickson, T. J. Am. Chem. SOC.1990, 112, 6127. (4) (a) Cramer, C. J.; Truhlar, D. G. J . Am. Chem. SOC.1991,113,8305831 1,8552-8554. (b) Cramer, C. J.; Truhlar, D. G. Science 1992,256,213217. (c) Cramer, C. J.; Truhlar, D. G. J . Comput. Chem. 1992, 13, 10891097. (d) Cramer, C. J.; Truhlar, D. G. J . Comput.-Aided Mol. Des. 1992, 6, 629-666. (5) (a) Wong, M. W.; Frisch, J. J.; Wiberg, K. B.J. Am. Chem.Soc. 1990, 113,4776; 1991,114,523. Wong, M. W.; Wiberg, K. B.; Frisch, J. J. J. Am. Chem. SOC.1992, 114, 1645. (b) Gould, I. R.; Green, D. V. S.; Young, P.; Hillier, I. H. J . Org. Chem. 1992,57,4434. (c) Ford, G. P.; Wang, B. J . Am. Chem. SOC.1992, 114, 10563-10569. (d) Wang, B.; Ford, G. P. J. Chem. Phys. 1991, 97, 4162-4169. (6) Hermann,R. B. J . Phys. Chem. 1972,76,2754. Kang,Y. K.;Nemethy, G.; Scheraga, H. A. J . Phys. Chem. 1987, 91, 4105, 4109, 4118. OOi, T.; Oobatake, M.; Nemethy, G.; Scheraga, H. A. Proc. Nutl. Acad. Sci. U.S.A. 1987,84, 3086. (7) The Merck Index: Budiavari, S., Ed.; Merck and Co., Inc.: Rahway, NJ,' 1989. (8) (a) Jacobsen, P.; Labouta, I. M.; Schaumburg, K.; Falch, E.; Krogsgaard-Larsen, P. J . Med. Chem. 1982,25,1157-I 162. (b) KrogsgaardLarsen, P.; Thyssen, K.; Schaumburg, K. Acta Chem. Scand. 1978, 832, 327-334. (c) Brehm, L.; Krogsgaard-Larsen, P.; Johnston, G. A. R.; Schaumburg, K. Acta Chem. Scand. 1976,830, 542-548. (d) KrogsgaardLarsen, P.; Johnston, G. A. R. J . Neurochem. 1975, 25, 797-802. +
l,a=eq b = ax
..
2
3 ,* = -,H
4 , * = -,H
* = -,H
cq5
HiF 6
7
8
In 1 N DCl/D20 (1, * = H) the conformer population shifts to 60-70% equatorial ( ~ J H . ~=~8.8 ~ /Hz), H . whereas ~ in base (1 N NaOD/D20) 1 is converted to the free-amine carboxylate ion, 5, present to the extent of 85-95% as the equatorial isomer (jJH.2axlH.3 = 11.1 Hz). In sharp contrast, the proton NMR spektrum of 3-fluoropiperidine hydrochloride, 2,lO in D2O reveals the presence of a single conformer with C(3)-F occupying an axial orientation. The latter was assigned by analysis of chemical shifts and coupling constants, the key diagnostic couplings occurring between H-3 and H-2ax, H-2, (3JH.2ax/H-3 = 1.6 Hz and 3 J ~ . 2 , 1 ~ . 3= 4.5 Hz, respectively) and between F(3) and H-2,,, H-2,, H-4,,, and H-4, (3J~.2ax,~ = 38.7, 3 J ~ . 2 c q ,=~ 9.7, 3JH.4ax,F = 44.5, and 3JH.4ep/H.3 = 9.3 Hz, respectively, 619F = -250.8).11 Dissolved in 1 N NaOD/D20,2 is converted reversibly to free-base 6 ('JH.3.F = 48 Hz).I2 The compound appears to exist entirely as the ring-inverted equatorial conformer. The I3CNMR spectrum of 6 displays long-range I9F coupling to C-5 (3Jc.5,F= 4.5 Hz) consistent with equatorialorientationof fluorine. This three-bond long-range coupling is completely absent in the C- 13 spectrum of the corresponding HCl salt 2 where fluorine is axial.I3 The NMR spectra for 3-carboxy-3-fluoropiperidine, 3,"J substantiate the presence of a single common F-axial conformation in D20 (zwitterion, = -), 1 N DCl/D20 (monocation, * = H), and 1 N NaOD/D20 (free-base 7) by virtue of the magnitude of the unchanged 1 H and 19F coupling constants (e.g., zwitterion: 3JH.2ax,F = 34.3 Hz, 3JH.2q,~ = 10.2 Hz, 3JH.4ax,F = 44.4 Hz, and 3 J H . 4 q , F = 10.6 Hz; 6'9F = -164.3). Finally, the proton NMR spectra of 3-carboxy-5-fluoropiperidine, 4,1°in D20 (zwitterion, * = -) or in 1 N DCl/D20 (monocation, * = H) are essentially the same. Examination of the IH, 'H couplings (3J~.j/~.2,x = 3.6 Hz, 3 J ~ . 3 / ~ . 2 q= 3.0 Hz) and the 'H, 19Fcouplings ( 3 J ~ . 6 a x=/ ~ 35.5 Hz in D20 and 37.5 Hz in 1 N DCl/D,O; 3 J H - 4 a x / ~= 44.4 (9) (a) Abraham, R. J.; Medforth, C. J.; Smith, P. E. J. Comput.-Aided Mol. Des. 1991.5.205-212 and references cited therein. (b) Terui. Y.; Tori, K. J. Chem. SOC.,Perkin Trans. 2 1975, 127-133. (10) (a) The fluoropiperidines 2 4 were prepared from the corresponding protected alcoholsusing (diethy1amino)sulfurtrifluoride (DAST)'Ob,cfollowed by deprotection. Synthetic schemes are found in the supplementary material. All new compounds gave satisfactory elemental compositions and/or highresolution mass spectra and were fully characterized by ' H and I3C NMR spectra; complete experimental details will be provided in the full paper. (b) Wilkinson, J. A. Chem. Rev. 1992,92,505-637. (c) Hudlicky, M. Org. React. ( N . Y . ) 1988, 35, 513-637. ( 1 1) (a) Williamson, K. L.; Li Hsu, Y.-F.; Hall, F. H.; Swager, S.; Coulter, M. S. J. Am. Chem. SOC.1968,90,6717-6722. (b) Williamson, K. L.; Li, Y.-F.; Hall, F. H.; Swager, S. J. Am. Chem. SOC.1966,88, 5678-5680. (1 2) The free base 6 has previously been prepared by an alternate synthetic route and reported to show 2 J ~ .F1= 48 Hz. See: Alvernhe, G.; Laurent, A,; Touhami, K. J. Fluorine Chem. 1985, 29, 363-384. (13) Indeed, thecalculateddistancebetweenC(3)-FandC-5 in2approaches 3 A, and three-bond through-space couplings of this nature are not expected. For a discussion of this point, see: (a) Myhre, P. C.; Edmonds, J. W.; Kruger, J. D. J. Am. Chem. SOC.1966,88, 2459. Holland, H. L. Teirahedron Lert. 1978, 881-882. (14) Frisch, M. J.; Trucks, G. W.; Head-Gordon, M.; Gill, P. M. W.; Wong, M. W.; Foreman, J. B.; Johnson, B. G.; Schlegel, H. B.; Robb, M. A.; Replogle, E. S.; Gomperts, R.; Andres, J. L.; Raghavachari, K.; Binkley, J.S.;Gonzalez,C.;Martin,R.L.;Fox,D. J.;Defrees,D. J.;Baker, J.;Stewart, J. J. P.; Pople, J. A. Gaussian 92, Revision A; Gaussian, Inc.: Pittsburgh, PA, 1992.
OOO2-7863/93/1515-3356$04.O0/0 0 1993 American Chemical Society
J. Am. Chem. SOC.,Vol. 115, No.8, 1993 3357
Communications to the Editor Table I.
Zwitterion and Piperidinium Salt Conformer Populations and E,,[ and E,,,, (298 K, kcal/mol) populations, 7% exptl NMR" H
F
PRDDOh
solvation E's AMSOL
'HzO"
gas phase AMBER'
AMSOL'
AMBER/FEPdf
AGo(aq)
AAGo(aq)
IS
0 ax 7% eq AE
100
30 70
36.2
100 0 10.7
16 84 -1 .o
93 7 1.5
0.4 99.6 -3.2
0
-0.5
-37.2 -51.8
14.6
1.8
2 % ax % eq
100
0
96 4 -1.9
99.6 0.4 -3.3
99.9 0.1 -4.3
99.6 0.4 -3.3
87 13 -1.1
82 18 -0.9
-57.2 -59.0
% axt % eq AE
97 3 -2.1
96 4 -1.9
0
0 100 8.8
99 1 -2.8
38 62 0.3
99.8 0.2 -3.8
-51.3
100 29.5
% ax % eq AE
93 7 -1.5
96 4 -1.9
100 0 -14.6
99.7 0.3
95
AE
3 -38.8
-12.5
-41.1 -52.9
11.8
4
100
0 -32.1
-3.5
5 -1.8
100
0 -6.2
Experimental ratios were obtained by the J-value method using extreme values 3J;Ia= 12.2 Hz,)JCc= 3.0 Hz for proton-proton coupling (cf. ref 9) and IJJ4= 46.0 Hz, 3Jcu= 12.0 Hz for proton-F coupling (cf.: Berthelot, J.-P.; Jacquesy, J.-C.; Levisalles, J . Bull. SOC.Chim. Fr. 1971, 1896). The F-derived quantities can be regarded as a lower limit. The J-value method generally provides populations within &5%. Populations are Boltzmann at 299 K. Halgren, T.A.; Kleier, D. A.; Hall, J . H., Jr.; Brown, L. D.; Lipscomb, W. N. J . Am. Chem. SOC.1978, 100, 6595. Throckmorton, L.; Marynick, D. S.J . Compur. Chem. 1985,6,652. E Reference 3; The N+-C-C-F torsional potential was derived with the 6-3 I IG* basis set in Gaussian 92 (ref 14) for use in MacroModel 3.5a (ref 3); v 1 / 2 -1.3405, V2/2 = -0.3166, V3/2 = 0.5709. For AMBER these values apply for phase angles of y I = 73 = 0.0, and y~ = 180.0. Each of the structures was supplemented with STO-3G electrostatic potential derived charges obtained by the Merz-Singh-Kollman scheme as implemented in Gaussian 92 (ref 14). Reference 4. / A M B E R , ref 2c; STO-3G E S P charges; 6-31 +G* ESP charges for I b and 2 yielded A E s of -3.0 and - 0 . 8 kcal/mol.
Hz in D2O and 1 N DCl/D20; 6 1 9=~ -183.4 in D20) establishes that both the fluorine and the carboxyl group are almost exclusively axial. Under basic conditions (1 N NaOD/D20) zwitterion 4 is transformed reversibly to the free amine carboxylate displaying exclusively the diequatorial form 8. This was confirmed in the I3C N M R spectrum of 8, as for 6, by the presence of long-range three-bond coupling between C(S)-F and C-3 ( 3 J ~ . 3=, ~6.0 Hz; 6lqF = -177.4). The unusual conformational variation within the series 1-4 can be understood primarily as a delicately balanced competition between electrostatic and solvation effects. Axial piperidinium species l b and 2 are strongly favored in the gas phase (Table I) as a consequence of unmitigated charge-charge (NH+- - -C02-) and charge-dipole (NH+- - -F*--C*+) attraction, respectively. AMSOL aqueous free energies of solvation (Table I) for 2 transferred to aqueous solution suggest a moderate AAG,O(axeq) = 1.8 kcal/mol relative stabilization for equatorial F, insufficient to overcome the axial Coulombic effect. For nipecotic acid, 1, however, AMSOL posits eq-CO2- to experience a solvation stabilization of nearly an order of magnitude greater than ax-CO2(AAG,O(ax+q) = 14.6 kcal/mol) in spite of the method's underestimate of the equatorial population. Application of this solvation estimate to the observed AAG(ax+q) = 0.5 kcal/mol for 1(* = -) implies that differential solvation and intramolecular charge neutralization effects (ca. 15 kcal/mol) nearly cancel. Placement of both F and C02- a t C-3 (3) maximizes the stabilizing factors for the ax-F/eq-COz- conformation as is observed (90-100%). In the case of 3,5-substitution, it is noteworthy that hypothetical addition of fluorine y and syn to the C02- in the predominantly equatorial 1 promotes axial repositioning of both moieties on going to 4, whereas similar addition of C 0 2 to fluoropiperidine 2 likewise giving 4 causes no conformational reorganization. The clear-cut stereochemical preference argues strongly that the axial (N-H)+- - -F*--C*+ charge-dipole association rather than the NH+- - -C02- ionic interaction tips the conformational balance.
The unique blend of steric, Coulombic, and solvation effects evident within compounds 1 4 a n d their neutraland singlycharged counterparts recommends series 1-8 as a challenging test suite for the qualitative and quantitative predictive ability of aqueous solvation treatments. For this reason we haveemployed a number of force fields (MM2, AMBER), charges (Mulliken, ESP), and quantum mechanics methods to predict the N M R populations for 1-8. A representative sample for 1-4 is given in Table I. While gas-phase ax/eq ratios do not model experiment, the predicted 'solution" values offer substantial improvement. The MacroModel AMBER GB/SA treatment draws closest to experiment when supplemented by an accurate N+-C-C-F torsional potential and ESP charges (Table I), but misrepresents the ax/eq populations of 2 and 4 by overestimating the stability of the diaxial zwitterions by 1.4 and 2.0 kcal/mol, respectively. The MM2 GB/SA variant with a refined N+-C-C-F torsion performs less satisfactorily. Given the lack of specific hydrogen bonding in these models, the level of agreement with experiment is encouraging. First solvation shell refinements within a continuum treatment might well eliminate the discrepancies. Surprisingly, the CPU-demanding AMBER free energy perturbation treatment with 650 explicit TIP3P water molecules matches experiment for 1 4 with less accuracy than thecontinuum models. For the conjugate acids (1,3,4, = H) and bases 5-8, the AMSOL aqueous models provide semiquantitative agreement with the N M R data (cf. supplementary material). In summary, the novel diaxial zwitterion 4 ( * = -) completes the deceptively simple series 1-4. The position of the chairchair conformational equilibria is recommended as a stringent test of the predictive ability of emerging treatments of aqueous solvation. Supplementary Material Available: Table of neutral and conjugate acid/base salt conformer population for 5-8 and synthetic schemes for 2-4 (4 pages). Ordering information is given on any current masthead page.
